MOAA, ET AL.-RT-1
BEFORE THE
POSTAL RATE COMMISSION
WASHINGTON, D.C. 202684001
POSTAL RATE AND FEE CHANGES,
)
2000 )
1
Docket No. R2000-1
REBUTTAL TESTIMONY
OF
22
2,
E
c-a
;$
2 ;.z
3,:
cor
-I.
i, -r
,a-.~
+-,.,
3.:~
=
r
5;
ROGER C. PRESCOTT
On B&half Of
MAIL ORDER ASSOCIATION OF AMERICA
DIRECT MARKETING ASSOCIATION, INC.
Communications with respectto this documentmay be sent to:
David C. Todd
Patton Boggs, L.L.P.
2550 M Street, N.W.
Washington, D.C. 20037-1350
(202) 457~6ooo
Counsel for Mail Order
Association of America
Due Date: August 14, 2000
z
$
5
0” t)
-75
=
-
ifs
TABLE OF CONTEm
I.
II.
INTRODUCTION
I!AGE
.. ............. ........... ...... ...... ... 1
PURPOSEOFTESTIMONY
...... ......... ......... ...... .... 4
SUMMARY
IV.
USPS STUDIES OF THE
RJZLATIONSHIP OF COSTS AND WEIGHT
........ ,..... ....... . 8
REVISED ANALYSIS OF THE
RELATIONSHIP
OF COSTS AND WEIGHT
. . . . . . . . . . . . . . . . . . . . . . 10
V.
VI.
AND CONCLUSIONS
........ ........ ........ ..... 6
III.
A.
Revised Cost Study Approach
B.
Statistical Reliability of RestatedRegressions .
C.
Impact of Statistical Outliers
D.
summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ...15
WITNESS HALDI’S AND
WITNESS TYR’S CRITIQUE
,
.
.
..
.
.
10
13
.
14
OF USPS’ STUDY . . . . . . . , . . . . . . . . . . . . 18
19
A.
Witness Daniel’s Changesto the Methodology Used in R97-1
B.
Witness Daniel AddressedPrevious Criticisms ..........
C.
Differences BetweenUnreliable and Imprecise Data ......
23
D.
Adjustments to Reflect Worksharing ................
24
E.
Impact of Data for 15-16 Ounce Weight Interval .........
25
F.
Impact of Weighted Data ........................
27
G.
Impact of Thin and Non-Mail Handling Tallies ..........
29
H.
Summary ..................................
33
.
20
- 111-
LE OF CONTENTS
VII.
VIII.
COST DIFFERENTIAL
FOR LETTERS AND FLATS
PAGE
...............
.34
A.
Withdrawal of the DAL Adjustment ..........................
36
B.
Witness Haldi’s Methodology for Heavy Letters ..................
1. Comparison of USPS’ Procedure .........................
2. Conceptual Error ...................................
37
37
38
C.
USPS’ Use of Correct Data ..............................
D.
summary ...........................................
COST COVERAGE
FOR STANDARD
(A) AND FIRST CLASS MAIL.
.43
.
...
.45
A.
Changesin First Class Presort Cost Coverage ...................
46
B.
Changesin ECR Cost Coverage ............................
48
C.
Contribution to Institutional Costs ...........................
50
D.
Summary .........................................
..5 1
-iv-
LIST OF EXHltUTS
EXHIBIT
(1)
TI LE
(2,’
MOAA, ET AL.-RT-1A
USPS’ Costs by Volumes and Weight Increments--Regular
MOAA, ET AL.-RT-1B
USPS’ Costs and Volumes by Weight Increments--ECR
MOAA, ET AL.-RT-1C
RestatedRegressionUtilizing Cost Per Pound and PiecesPer
Pound
MOAA, ET AL.-RT-1D
USPSWitness Daniel’s Estimated Standard(A) ECR Unit Cost
Per Piece
MOAA, ET AL.-RT-IE
Unit Cost Line Basedon RestatedRegression--Regular
MOAA, ET AL.-RT-IF
Unit Cost Line Basedon RestatedRegression--ECR
MOM,
ET AL.-RT-1
I. INTRODUCTION
1
My name is Roger C. Prescott. I am an economist and Executive Vice President of the
2
economic consulting firm of L. E. Peabody& Associates,Inc. The firm’s offices are located at
3
1501 Duke Street, Suite 200, Alexandria, Virginia 22314. I am the sameRoger C. Prescott who
4
previously submittedDirect Testimony in this proceedingon May 22, 2000 on behalf of the Mail
5
Order Association of America (“MOAA”).”
6
my Direct Testimony
My qualifications were attachedas Appendix A to
7
In this current proceeding,PostalRateCommission(“PRC”) Docket No. R2000-1, PostalRate
8
and Pee Chancres.2ooQ(“R2000-I”), the United StatesPostal Service’s(“USPS”) Witness Sharon
9
Daniel (USPS-T-28)submitteda study which examinesthe impact of changesin weight on changes
10
in unit costsfor Standard(A) mail. The results of Witness Daniel’s study were used, in part, to
11
support the USPS’ proposedrate structurefor Standard(A) mail which included a decreasein the
12
pound portion of the piece/poundrate for mail weighing greaterthan 3.3 ounces(i.e., mail above
13
the breakpoint). Specifically, in this proceeding the USPS is proposing to decreasethe pound
14
portion of the piece/pound rates as shown in Table 1 below.
u
I also submitted Direct Testimony in this proceeding on behalf of E-Stamp Corporation
-2-
4
5
IT!2
-
6
1.
7
8
MOAA,
ET AL.-RT-1
Table 1
Summary of Current and USPS’ Proposed
:.
3
9
Per Pound Portion of
PiecelPound RateI,
Item
lie&arm
(2)
(3)
current
67.7C
66.3C
2.
As hoposed by USPS”
!5.6&
ze!d.s
3.
Difference (Lie 2 Lie
(-)1.6C
(-)7.9C
(1)
1)
3
-
1:
11
12
Y
For mail weighing greater than 3.3 ounces per piece.
Excludes discounts for destination entry.
Moeller (USPS-T-35), pages 8 and 19.
13
As shown on Line 3 of Table 1 above,the USPSis proposingto decreasethe pound portion of the
14
piece/poundrate by 1.6 centsper pound for Standard(A) Regular (“Regular”) mail and 7.9 cents
15
per pound for Standard(A) EnhancedCarrier Rate (“ECR”) mail.
16
In addition to the abovechangesto the per pound portion of the Standard(A) rate structure,
17
the USPShas also proposedcontinuationof the differencesin the per piece ratesfor letter-shaped
18
and flat-shaped ECR mail qualifying for the high density and saturation discounts. The USPS’
19
proposedrate differential for letters and flats equals0.2 cents per piece for high density mail and
20
0.5 centsper piece for saturation mail.”
z,
Moeller, page 28, (as summarized in Table 7 to this Rebuttal Testimony)
-3-
MOAA. ET AL.-RT-1
1
Finally, the USPS has presentedrates that reflect a cost coverageratio of 209 percent in its
2
Test Year After Rates (“TYAR”) analysis.“’ This coverage ratio is 24 percent higher than the
3
USPS’ proposedcoverageratio of 168 percent for all mail and services.
3
Mayes (USPS-T-32). Exhibit-32B
-41
II. PyBpoSE
MOM,
ET AL.-RT-1
OF TESUMONy
2
I have been askedby MOAA and Direct Marketing Association, Inc. (“MOAA, et al.“) to
3
review the Direct Testimony, the responsesto interrogatories and the underlying workpapers of
4
Val-Pak Marketing Systems, Inc., Val-Pal Dealers’ Association, Inc. and Carol Wright
5
Promotions, Inc.‘s (referred to collectively herein as “VPKW”) Witness John Haldi and
6
NewspaperAssociationof America’s (“NAA”) WitnessWilliam B. Tye in order to evaluate.their
7
respective testimony related to the USPS’ Witness Daniel’s study of the changesin costs when
8
weight changes. In order to addresstheir criticisms, I have reviewed Witness Daniel’s testimony
9
(and supporting workpapers) in order to assessthe validity of her conclusion regarding the
10
existence of a relationship between changesin mail weight and changesin the unit costs for
11
Regular and ECR mail. I have also been requestedby MOAA to review VP/CW’s Witness
12
Haldi’s assertionthat within ECR mail, the USPShas systematicallyoverstatedthe costsof letter-
13
shapedmail while correspondinglyunderstatingthe costsof flat-shapedmail. Finally, I havebeen
14
asked by MOAA to review the testimony of American Bankers Association and National
15
Association of PresortMailers’ (referred to collectively herein as “ABA/NAPM”) Witness James
16
A. Clifton regarding the fairness of the coverageratio of Standard(A) mail versus First Class
17
mail.
18
The results of my analysesare summarized under the following topics:
19
III.
Summary and Conclusions
20
IV.
USPS Studiesof the Relationship of Costs and Weight
21
V.
Revised Analysis of the Relationship of Costs and Weight
-5
1
VI.
Witness Haldi’s and Wimess Tye’s Critique of USPS’ Study
2
VII.
Cost Differential for Letters and Flats
3
VIII.
Cost Coveragefor Standard(A) and First Class Mail
MOM,
ET AL.-RT-1
-61
III.
SUMMARY
MOAA,
ET AL.-RT-1
AND CONCLUSIONS
2
Basedon my review and analysis of the testimony presentedby WitnessesHaldi and Tye, I
3
conclude tbat the per pound portion of the rates for Standard(A) Regular and ECR mail should
4
be no higher than the rates proposedby the USPS. I also concludethat no basis exists to support
5
Witness Haldi’s contention that the cost differential between letter-shapedmail and flat-shaped
6
mail developedby the USPSshould be increased. Finally I conclude that the proposedcoverage
7
ratio for ECR mail is not improper when comparedto First Classmail. My conclusionsare based
8
on the analysespresentedin this Rebuttal Testimony and summarizedas follows:
9
10
11
12
1. The PRC should acceptthe rates for Standard(A) Regular and ECR mail as presentedby
the USPS’ WitnessMoeller (USPS-T-35). As part of his rate structure, Witness Moeller
proposed a pound rate for mail weighing more than 3.3 ouncesof 66.1 centsper pound
for Regular mail and 58.4 centsper pound for ECR mail.
13
14
15
16
17
18
19
2. Using the samebase data as relied upon by Wimess Daniel, an alternative simple linear
regressiondemonstratesa strong relationship betweenchangesin the cost per pound and
changesin the number of piecesper pound. For Regular mail, my regression identifies
a cost line equal to the sum of 11.1 centsper piece plus 52.5 cents per pound. For ECR
mail, my regressionidentifies a cost line equal to the sum of 5.6 centsper piece plus 17.6
cents per pound. This analysis supportsthe conclusion that the per pound portion of the
rates should be no greater than the ratesproposedby the USPS.
20
21
22
23
3. Witness Haldi’s and Witness Tye’s criticism regarding the lack of support for a reduced
pound rate for Standard(A) mail is misplaced. When their criticisms are viewed in light
of my analysis, the per pound portion of the rates should be no higher than the USPS’
proposed rates for Standard(A) mail.
24
25
26
27
28
29
30
4. Witness Haldi is incorrect in his claim that a mismatchoccursbetweenthe costs for ECR
letters andECR flats. WitnessHaldi’s cost calculationsare basedon improper procedures
which do not measurethe actual cost differences betweenletters and flats qualifying for
the ECR high density or saturation rates. In addition, in the sourcedata relied upon by
Witness Haldi, the USPShasalready correctedthe data to considerany mismatch. When
the data is properly analyzed,the USPS’ calculation of a letter/flat differential of 0.2 cents
per piece for high density mail and 0.5 cents per piece for saturation mail is valid.
-7-
MOM,
ET AL.-RT-1
5. Witness Clifton’s assertion that ECR mail is receiving preferential treatment when
compared to First Class mail is not valid. Much of the increasethe coverageratio for
First Classmail is due to decreasesin costswhich the PRC has recognized as support for
increasing the cost coverage. The contribution per piece for ECR mail has increasedat
the sameapproximate rate as First Class mail. In addition, averagerates for ECR mail
have increasedalmost twice as fast as First Class mail over the time period studied by
Witness Clifton.
4%
1
IV. <
MOM,
ET AL.-RT-1
OF COSTS AND WEIGHT
In the past, the USPS has conductednumerous studies regarding the impact of changesin
costs causedby changesin weight. These studies date as far back as Witness Madison’s study
in PRC Docket No. R84-1, Postal Rateand Fee Cm
(“Docket No. R84-1”). The USPS
submitted a similar study in PRC Docket No. MC95-1, Mail Classification Schedule. 1995,
Classlficatlon
(“Docket No. MC95-1”). The most recent study submitted to the PRC
prior to the presentproceeding was by the USPS’ Witness Michael R. McGrane in PRC Docket
8
No. R97-1, PostalRate -es.
1997(“R97-1”). WitnessMcGrane concludedthat unit
9
costshave an upward trending relationshipwith mail weight. *’ The PRC, in its Docket No. R97-1
10
decision, criticized Witness McGrane’s study for excluding a comprehensiveanalysis of cost-
11
causingfactors. Specifically, the PRC questionedWitnessMcGrane’sassignmentof non-In-Office
12
Cost System(‘TOCS”)related coststo weight incrementsbasedupon various volumetric measures
13
rather than using more appropriate weight-basedmetrics. u
14
The USPS’ current study of the impact of changesin costsdue to changesin weight submitted
15
by the USPS’ Witness Daniel addressedWitness McGrane’s non-IOCS cost assignment. In her
16
analysis Witness Daniel allocated the elemental load portion of street delivery costs based on
17
weight by shapeinstead of on pieces as was done by Witness McGrane in Docket No. R97-l!’
18
Witness Daniel also refined her study by improving the methodology utilized to distribute mail
51
2
4i
Docket R97-1, McGrane, (USPS-ST-44) Exhibit USPS-44B, page 2.
Docket R97-1 decision, page 401.
Daniel, page 9.
-9-
MOAA, ET AL.-RT-1
1
processing volume variable costs to weight increments.” In my opinion, Witness Daniel’s new
2
technique improves on the methodology employed by the USPS’ Witness McGrane in PRC
3
Docket No. R97-1, as well as the studiespresentedin Docket No. MC951 and Docket No. R84-
4
1.
5
In responseto interrogatories submitted by ADVO, Witness Daniel refined her model to an
6
even greaterextent. ADVO requestedthat WitnessDaniel provide costs, mail volumes, and unit
7
costsfor ECR in total and ECR flats with adjustmentsto eliminate the cost savingsassociatedwith
8
worksharing (i.e., the cost savings for destination entry and presortation)s The results of
9
ADVO’s requestis an analysiswhich providesmore information aboutthe causativefactors of the
10
cost-weight relationship than any previous USPS study.s’
?I
81
9,
The details of Witness Daniel’s study were presented in Library Reference USPS-LR-I-92 (“LR-92”) and the
responses to ADVO, Inc.‘s (“ADVO”) interrogatories.
ADVOKISPS-T28-10 through ADVOIUSPS-T-28-11.
As discussed below, I have not relied on the additional detailed information provided in response to ADVO’s
interrogatories to develop my restated regressions because my analysis shows a significantly bigh correlation of
cost to weight utilizing all costs for each weight interval (including worksharing reductions). No futber
adjustment was necessary to show the validity of the USPS’ proposed pound portion of the piece/pound rates.
-lO-
.
MOAA,
ET AL.-RT-1
1
V. REVISED
2
Witness Daniel’s approachto analyzingchangesin costs as weight changesproduced results
3
that were subjectto criticisms raisedby WitnessesHaldi and Tye. Theseare discussedin Section
4
VI below to this Rebuttal Testimony. In summary, using a more appropriate approach to
5
analyzing the relationship of cost and weight while still relying on Witness Daniel’s base data
6
demonstratesthat the IOCS producesvalid data which can be reliably used to show the effect of
7
weight upon costs. This sectionof my RebuttalTestimony presentsmy alternative approachand
8
the results of my analysis under the following topics:
9
ANALYSIS
OF ‘lXU&I3LATIONSH
A. Revised Cost Study Approach
10
B. Statistical Reliability of RestatedRegressions
11
C. Impact of Statistical Outliers
12
D. Summary
13
14
WEIGHT
A.-
Witness Daniel implicitly utilized the cost relationship between five interacting elements to
15
derive her unit costs for Regular and ECR mail. The relationship of changesin costs associated
16
with changesin weight for Standard(A) mail can be viewed as the interaction of the following five
17
key factors:
18
1. Volume;
19
2. Weight;
20
3. Aggregate costs;
-ll-
MOM,
ET AL.-RT-1
1
4. Average Unit Cost Per Piece; and
2
5. Average Unit Cost Per Pour@
3
WitnessDaniel utiliied the averageweight per piece and averagecost per piece as inputs into
4
her regressionmodel. 11’ In other words, shedevelopedthe averageweight per piece (for each
5
interval) by dividing total weight for that weight interval by total pieces for the weight interval.
6
Next, she divided the aggregatecosts for each weight interval by total pieces for the weight
7
interval to develop average costs per piece. The cost per piece and averageweight per piece
8
utilized as the inputs for Wimess Daniel’s regressionmodel for Regular mail are summarized in
9
Cohmm (5) and Column (6) of Exhibit MOAA, ET AL.-RT-1A to my Rebuttal Testimony. The
10
inputs for Witness Daniel’s regressionmodel for ECR mail are summarized in Column (5) and
11
Column (6) of Exhibit MOAA, ET AL.-RT-1B to my Rebuttal Testimony. Witness Daniel’s
12
approachresults in statistical outliers as well as wide variancesin weight and volume as pointed
13
out by WitnessesHaldi and Tye.
14
Another approach to studying the cost-weight relationship, and one that I applied in my
15
analysis, is to determine the averagecost per p,o,undrather than the averagecost per p&e as
?p/ From a mathematical perspective, the cost-weight relationship can be described by the following equation:
Y
where
II’
Y
Xl
X,
=
a*x,+b*x,
= Total cost within a weight interval
= Total volume @ieces) within a weight interval
= Total weight within a weight interval
= The average unit cost per piece
= The average unit cost per pound
t
Algebraically, Witness Daniel divided the factors in the equation above by the total number of pieces. This
process yields the following equation: YIX, = a + b(X,/X,) which is equivalent to the regreSSiOnliieS in
Witness Daniel’s analysis in LR-92.
-12-
jl
MOAA, ET AL.-RT-1
1
WitnessDaniel has done in her model.JZ!Instead of dividing the equation by the total number of
2
pieces, I divide by the total pounds. The inputs for my analysis utilize the averagepieces per
3
pound (16 ouncesper pound + the averageweight per piece for each weight interval) and the
4
averagecost per pound (total cost per weight interval i total pounds per weight interval).
Using the Regular mail data containedin Exhibit MOAA, ET AL.-RT-1A and the ECR mail
5
6
data containedin Exhibit MOAA, ET AL.-RT-lB, I performed a simple least squaresregression
7
to determine the averagecost per piece and averagecost per pound for both Regular and ECR
8
mail. For Regular mail, my regressionutilizes the piecesper pound and the cost data shown in
9
Column (7) and Column (8) of Exhibit MOAA, ET AL.-RT-1A. For ECR mail, my regression
10
utilizes the pieces per pound and the cost data shown in Column (7) and Column (8) of Exhibit
11
MOAA, ET AL.-RT-1B. The results of my regressionanalysesare shown in Table 2 below.
Table 2
&nnna~
of
Reamsalan
Results
(2)
Cents Per Piece
(3)
Subclass(1)
R Sauare
(4)
16
1.
Regular
52.5
11.1
0.959
17
!.
ECR
17.6
5.6
0.965
18
kwrces: Spreadsheet titled “Prescott workpapers for MOAA, ET AL-RT-l.xls”
submitted with this testimony.
iii
U
My analysis derives the following equation: Y/X, = b + a (X,/X,). Then using algebra, it is possible to cmml
the cost per pound equation to an equivalent cost per piece equation.
-131
2
MOM,
ET AL.-RT-1
My regression analysis producesthe following equationsto calculate the cost per piece as
weight changesfor Regular and ECR mail:
3
4
Regular
= 11.1 centsper piece + [52.5 centsper pound x (ouncesper piece + 16 ounces
per pound)]
5
6
ECR
= 5.6 centsper piece + [17.6 centsper pound x (ouncesper piece + 16 ounces
per pound)]
7
Exhibit MOAA, ET AL.-RT-1C is a graphical comparison of the data points and my
8
regressionanalysesfor Regular and ECR mail.U’ The results of my analysis above show that the
9
cost per pound for Regular mail is much larger than the cost per pound for ECR mail. This is not
10
unexpectedand, in fact, was recognizedby WitnessHaldi. In his testimony, WitnessHaldi noted
11
that ECR mail cost less than Regular mail. 14,In addition, the greater presortation and depth of
12
dropshipping also contribute to reductions in weight related costs for ECR mail.
13
14
B.
STATISTICAL
RELIABILITY
OF RJZSTBTED RBGIW&IQNS
15
The reliability of the results of a regressionanalysiscan be judged by various statistics. The
16
key statistic in a regressionanalysisis the coefficient of determination, or more commonly known
17
as the R-squared (“R’“) value. In my analysis, the R value illustrates the proportion of the
18
variability in the cost per pound which is explainedby the relationship to the number of piecesper
19
pound. In other words, how much of the changein the cost per pound is explainedby the change
20
in the number of pieces per pound.
u’
li’
Page 1 of 2 of Exhibit MOAA. ET AL-RT-1C graphically shows the data points and regression line for Regular
mail. The data points and regression lie for ECR mail are graphically depicted on Exhibit MOAA, ET AL.RT-lC, page 2 of 2.
Tr. 32/15759.
-14-
MOAA,
ET AL.-RT-1
1
As shown in Table 2 above,my revisedregressionhasan I? value of 95.9 percentfor Regular
2
mail and 96.5 percentfor ECR mail. The regressionin my analysisindicatesthat over 95 percent
3
of the change in the cost per pound is explained by changesin the pieces per pound. This
4
illustrates that there is a strong relationshipbetweenchangesin unit costs and changesin weight.
5
C. IMP AT
6
As I discussin Section VI. below, Witness Tye and Witness Haldi fault Witness Daniel for
7
the extreme outlying values in her data and the measuresshe took to improve her analysesby
8
combining weight intervals. Aggregating data, as Witness Daniel has done, minimizes the
9
negative impact of the outlying data, but this aggregation also hides important explanatory
10
information, A better methodology is to retain all the data, in as much detail as possible. My
11
restatedanalysispresentedhere in my RebuttalTestimony accomplishesthis goal,becausethe form
12
of the data I haveutilized (i.e., costsper pound and piecesper pound) maintains all of the weight
13
intervals, but does not result in any outlying values.
14
I have createda graphical exampleto demonstratethis point. Exhibit MOAA, ET AL.-RT-
15
1D represents the graph preparedby Witness Daniel illustrating the relationship she developed
16
betweenECR unit costsand weight per piece. Data Point A in Exhibit MOAA, ET AL.-RT-1D
17
is the averagecost per piece for mail within the 15-16 ounceweight interval. As Exhibit MOAA,
18
ET AL.-RT-1D illustrates, Data Point A shows a wide variance from the other values in the set
19
of dam utilized by Witness Daniel.
-15-
MOAA,
ET AL.-RT-1
1
In contrast to the data set used by Witness Daniel in her analysis, the data set I utilized
2
contains no statistical outliers. Exhibit MOAA, ET AL.-RT-lC, page 2 of 2 is a scatter-plot
3
diagram of the data range for ECR mail used in my revised analysisutilizing the samebasic data
4
from LR-92 (aggregatepieces, pounds and costs)that Witness Daniel utilized u’. Data Point A
5
in Exhibit MOAA, ET AL.-RT-1C showsthe averagepiecesper pound in the 15-16 ounceweight
6
interval. This is the same weight interval which produced the statistical outlier in Witness
7
Daniel’s analysis. In my revised analysis the data for the 15-16 ounce weight interval is within
8
the normative range of the entire data set. By examining the cost-weight relationship from this
9
revised approach, I have retained the explanatory value of each outlying weight interval but
10
eliminated the statistical abnormalities included in Witness Daniel’s data set.
11
D.WM&QQ!
12
My revised cost-weight study illustrates two points. First, the regression model in my
13
analysisof the USPS’ data showsa strong relationshipexistsbetweenchangesin costsand changes
14
in weight for Standard(A) mail. As describedabove, 95.9 percent of the changein unit costs in
15
Regular mail is explained by changesin the weight function. The results are equally significant
16
for ECR mail with 96.5 percent of the changein unit costs explained by changesin the weight
17
function.
18
Second, the practical implication of my revised analysis is an ability to generate, with
19
statistical accuracy, the estimated unit
20
intervals. Table 3 below summarizesthe cost line for Regular and ECR mail basedon the results
W
The numeric values for the
costs
inputsare shown in
for both Regular and ECR mail at key weight
Exhibit MOAA, ET AL:RT-1B.
-16-
MOAA,
ET AL.-RT-1
1
of my regressionanalysis for each 1 ounce increment and the rate breakpoint of 3.3 ounces. A
2
graphical representationsof theseestimatedunit costsare found in Exhibit MOAA, ET AL.-RT-
3
1E and MOAA, ET AL.-RT-1F to my Rebuttal Testimony.
Table 3
Estimated Unit Costs for
(cent.3per piece)
;
10
11
12
;:
:i
17
18
;z
z:
23
24
25
26
27
28
zi
31
32
Weight
ir2lu&
(1)
1. I
2. 2
3. 3
4. 3.3
5. 4
6. 5
7. 6
8. 7
9. 8
10. 9
11. 10
12. 11
13. 12
14. 13
15. 14
16. 15
17. 16
&g&d
(2)
14.4
17.7
20.9
21.9
24.2
27.5
30.8
34.1
37.3
40.6
43.9
47.2
50.5
53.7
57.0
60.3
63.6
jj(&
(3)
6.7
7.8
8.9
9.2
11.0
11.1
12.2
13.3
14.4
15.5
16.5
17.6
18.7
19.8
20.9
22.0
23.1
11.1 centsper piece + (52.5 centsper pound 16 ouncesx weight per piece in ounces).
5.6 centsper piece + (17.6 centsper pound f
16 ouncesx weight per piece).
33
The USPS has proposed rates with the pound portion equaling 66.1 cents per pound for
34
Regular mail and 58.4 cents per pound for ECR mail. My regression results in a pound
35
componentof the costsequaling 52.5 centsper pound for Regular mail and 17.6 cents per pound
36
for ECR mail. My regression analysis demonstratesthat the USPS’ proposed pound portion of
-17-
MOAA,
ET AL.-RT-1
1
the rates increasesfaster than the actualpound-relatedcosts.My analysis supports the conclusion
2
that the changesin the pound portion of the ratesproposedby the USPSas shown in Table 1 above
3
are justified and fair.
-1%
MOAA, ET AL.-RT-1
1
VI. WITNESS I-I&DI’S
2
Both VPICW’s Witness Haldi and NAA’s Witness Tye criticize the underlying data and
3
results of Witness Daniel’s study of the changesin costs with changes in weight. Many of
4
WitnessesTye and Haldi’s criticisms are simply rhetoric and unfoundedassertions. This doesnot
5
mean that the way Witness Daniel presentedher data and the results could not be improved (as I
6
demonstratedabove). Any perceived shortcomingsin Witness Daniel’s study can be overcome,
-I
as I have shown, by using simple statistical procedures.
8
9
AND WITNESS TYE’S CRITIOUE
OF USPS’ STUDY
In general, WitnessHaldi rejectsthe USPS’sstudy as a wholly inadequatetool for ratemaking
purposes, stating:
10
11
12
13
14
“...studies of the weight-cost relationship offered by the Postal Service in this
docket must again be rejected as inadequateto demonstratethat the effect of
weight on cost is overstated. They provide no basis for the commission to
recommend a drastic reduction in the pound rate as requested by the Postal
Service.“fi’
15
Witness Tye summarizeshis criticism of Witness Daniel’s weight-cost analysis, stating:
16
17
18
19
20
“I find that the data gatheredfor this analysis are not reliable. I further find that
the cost data are inconsistentlyapplied to justify Fist Classand Standard(A) rate
design proposals...Since the result of the distribution key analysis (weight-cost
study) are “cherry picked”, they form no reliable basis for changesin the ECR
pound rate” u’
Xi1 Tr. 32l1.5772.
12' Tr.30114692.
-19-
,
MOAA, ET AL.-RT-1
This sectionof my Rebuttal Testimony summarizesthe criticisms raised by WitnessesHaldi
and Tye and explains the reasonswhy I believe that their criticisms are not valid or that the effect
of the criticism can be overcomeby modifications to the statistical analysis of the data (as I have
done). My review is discussedunder the following topics:
5
A. Witness Daniel’s Changesto the Methodology Used in R97-1
6
B. Witness Daniel AddressedPrevious Criticisms
7
C. Differences Between Unreliable and Imprecise Dam
8
D. Adjustments to Reflect Worksharing
9
E. Impact of Data for 15-16 GunceWeight Interval
10
F. Impact of Weighted Data
11
G. Impact of Thin and Non-Mail Handling Tallies
12
H. Summary
13
14
A. WITNESS DANIEL’S CHANGES TO
p
IN R97-I
15
Witness Tye claims that the study prepared and submitted by Witness Daniel is nearly
16
identical to the study submittedby the USPS’WitnessMcGrane in Docket No. R97-1.!s’However,
17
Witness Daniel’s study made two (2) significant changesto the proceduresfollowed by Wimess
18
McGrane in Docket No. R97-1.
19
First, Witness Daniel changedthe allocation basis for the elemental load portion of street
20
delivery costs. In Docket No. R97-1, WitnessMcGrane allocatedtheseportions of elementalload
18’
Tr. 30/14698.
-2o-
..
MOM,
ET AL.-RT-1
1
costson a piece basis(elementalload includes the time spent handling mail pieces at the point of
2
delivery). le’ Studies have shown that shapeis a key driver in elemental load costs?’ Witness
3
Daniel, therefore, concluded that shapeis not the only force impacting elemental load cost, i.e.,
4
weight is also a pertinent factor of elemental load cost. a’
5
Second,WitnessDaniel modified the methodologyto distribute weight-basedcosts when the
6
actual weight may not be known. In the USPS’ weightcost study in R97-1, 22/the USPSused the
7
averagecost of mail for all subclassesto allocate unknown coststo each subclass. In her study,
8
Witness Daniel adoptedan alternativemethodology utilizing the information where the weight is
9
known within a specific cost pool, activity code or subclassto distribute cost information from
10
tallies where the weight is unknown. This use of costsassignedto each subclassto allocate costs
11
from tallies (for that specific subclass)with unknown weight provides a greater,level of precision
12
in the allocation of costs than was available in previous studies.?l’
13
14
B. WITNESS DANIEL ADDRESSED
PREVIOUS CRITICISMS
15
Witness Tye claims that Witness Daniel’s study fails to addressthe criticisms raised by the
16
PRC in Docket No. R97-1.24’He doesthis by selectivelyquoting portions of the PRC’s decision
17
in Docket No. R97-1 and by de-emphasizingthe changesmade by Witness Daniel in the study
18
presentedin this proceeding.
g
f,
22
&I’
Daniel, page 8.
Daniel, page 8.
Daniel, page 8.
McGram, USPS-ST-44.
Costs withii a specific cost pool, activity code or subclass provide a better proxy for allocating costs because
they more accurately reflect the characteristics inherent in that group of costs.
Tr. 3004698.
-21-
MOAA, ET AL.-RT-1
1
WitnessTye quotesthe PRC’s decision in Docket No. R!97-1in an attempt to show the USPS
2
had not respondedto the PRC’s criticisms of its previous study. z In the passagescited, Witness
3
Tye omitted portions of the PRC’s opinion which show that Witness Daniel did, in fact, address
4
the PRC’s criticisms of the Docket No. R97-1 study. Witness Tye’s selective quote showed two
5
criticisms which are, in fact, only one issueaddressedby the PRC, namely the assignmentof non-
6
IOCS costs (i.e., delivery costs) on a volumetric or piece basis instead of a weight basis.
7
Examination of the full text of the PRC’s opinion showsthat Witness Tye omitted the portions of
8
the PRC’s decision which are relevantto the evaluationof WitnessDaniel’s methodology. In the
9
text of the PRC’s opinion, Witness Tye did not include the underlined portions of the following
10
quotes:
11
12
13
“Another problem with the cost-weightstudy is that it containsno comprehensive
study of cost-causingfactors. LL&kzr-IOCS relatedcosts-Fried
to weight
Increment on the basis of various volumetric measures.,,z,
14
15
16
17
18
19
‘I
Q
i
ae niece-reI,&& Where the Servicehas failed to test these rationales or its own
theories, there is no soundbasison the record for distributing carrier street costs
to ounce increments. This is a serious shortcoming as elemental load time
accounts for approximately one-half of carrier street attributable cost.” 2zI
(emphasisadded)
20
21
As shown above, Witness Daniel changedthe method of allocating elemental delivery costs
in her study in this proceeding to addressthe PRC’s issuewith the earlier study.
2
Tr. 30114698.
R97-1 decision, page 401.
22’ R97-1 decision, page 402.
,
-22-
.
MOM,
ET AL.-RT-1
1
Witness Haldi believesthat the deficienciesin the IOCS data require that the USPSundertake
2
an engineeringstudy or a simulation analysisto gather the data necessaryfor an all encompassing
3
analysis of the impact of weight on costs.zsIWitness Tye also criticizes Witness Daniel’s use of
4
IOCS data.=’ Both Witness Haldi and Witness Tye overlook the benefits of utilizing IOCS data
5
and the shortcomings of that would be present in an empirical engineering study.
6
As Witness Daniel describesin her testimony, the IOCS supplied data provides an inherent
7
advantageover other potential data sourcesbecauseof the completenessof the data. She noted:
8
9
10
“An IOCS-basedanalysis, however, is adopted here becausethe IOCS samples
employeesin all mail processingand carrier in-office operationsaround the clock,
24 hours per day, 7 days per week.lcs’
11
In other words, the data supplied by the IOCS provides a view of the entire USPS operation and
12
not a limited subsetof individual postal processes.
13
14
Furthermore, Witness Daniel points out that a study of limited scope may not provide a
superior quality of data than that used in her analysis, noting:
15
16
17
It isdoubtful that a one-time field study could be superior to the data used in the
weight studiesdescribedin my testimony, which are basedon a national sample
of all operations over the course of a year?l
18
I agree with Witness Daniel that the use of the IOCS data representsthe broad spectrum of
19
the costs incurred by the USPS over an extendedperiod and should be superior to the data that
g
E
Tr. 3W1.5848.
Tr. 30/14700.
Daniel, page34.
Tr. 411174.
..
-23-
MOAA. ET AL.-RT-1
1
would be potentially developed in an engineering study of the changesin costs associatedwith
2
changesin weight. For purposesof this proceeding, the IOCS data is adequatefor me analysis
3
of changesin costs due to changesin weight as presentedby Witness Daniel and me.
4
5
C. DIFFERENCES BETWEEN
LIABLE AND IMPRECISE
6
DATA
Wimess Tye assertsthat Witness Daniel has discredited her own study, stating that “wimess
7
Daniel herself concedesmat her data are so unreliable as to be useful only for a broad view.“a’
8
Witness Daniel only qualified her testimony in regard to the precision of her study and not the
9
reliability of her analysis. Imprecision in the data doesnot mean that her study is unreliable.?
10
Contrary to Wimess Tye’s testimony, Witness Daniel does not concede that her data is
11
unreliable. WitnessDaniel only stipulatesthat her data is not precise, and in this instance, exact
12
precision is not necessary. In any analysis,the required precision of the data is a direct function
13
of the data’s end use. In other words, where less than exact data will suit the purpose of the
14
researcher,then exactprecision may not be necessary.Becausethe USPS has not used the study
15
as the only criteria for its proposedrates, further precision is not needed.
16
As Witness Daniel points out, the USPS’ pricing witnessesdo not utilize the exact weight
17
interval point estimatesproducedin her cost study for the USPS’ rate proposal. Instead of using
18
point estimates,the USPSusesthe overall cost relationshipsin their price setting analyses.N The
g
Tr. 30/14699.
In any event, my restatement of the study of changes in costs with changes in weight shows that a valid
relationship does exist and supports that the pound portion of the rate should be no higher than the rates proposed
by the USPS.
3~4’ Tr. 411307.
-24-
>
MOAA,
ET AL.-RT-1
1
overall conclusionof the regressionanalysesshowsthat the USPS’ rate structure is supported, in
2
total, by the resultsof the study of the changesin costsassociatedwith changesin weight. Stated
3
differently, the regression results are not used to set the pound portion of the proposed rates,
4
however, the regressionanalysis show, with great statistical reliability, that costs are increasing
5
much less rapidly than the USPS’ proposedrates.
6
7
D. ADJUSTMENTS TO
REFLECT WORKSHARING
8
Both WitnessHaldi and WitnessTye criticize Witness Daniel for improperly accounting for
9
worksharing. z’ When Witness Daniel doesadjust ECR flat data for worksharing, Witness Haldi
10
11
12
misstatesthe results of her adjustmentwhen he claims:
“An effort is made to adjust for destination entry which increases the weight
related cost over the initial effort.” (emphasisadded)%’
13
WimessHaldi’s conclusionis wrong. The exclusionof costsrelatedto worksharing decreases
14
the weight-related cost. Basedon the data utilized by Witness Daniel, Table 4 below compares
15
the estimated cost per ounce for mailing ECR flats before and after adjustments to reflect
16
worksharing; In all cases, the estimated weight-related cost per ounce in Witness Daniel’s
17
analysesdecreaseswhen worksharing adjustmentsare made.
lZz’ Tr. 3205827-15879 and Tr. 32/15838-15840
=’ Tr. 32115829.
-25-
.
1
2
MOAA,
ET AL.-RT-1
Table 4
Imoactof WorkshariaeAdiuslmen
t cmE&mated
.
CostPer Ounceof ECRFJ&
3
4
Item
(1)
CentsPerOunce
Worksharing
Difference4’
Unadiusted -21
(4)
(2)
(3)
5
1.
Utiiiing
6
2.
Utilizing Witness Daniel’s Combined Weight Increments
7
8
9
10
11
11
21
,r
”
LR-92, Worksheet LR92bECR.xls, sheet ECR Flats (detailed).
LR-92, Worksheet LR92bECRzls. sheet ECR Flats (combined).
ADVONSPS-T28-10 (Tr. 4/1210-1211).
c01umn (2) minus Column (3).
12
Witness Daniel’s Detailed Weight Increments
1.55c i’
1.42 C
0.13c
1.37u
1.24
0.13
In Witness Daniel’s data, worksharing reducesthe cost per ounce by 0.13 cents per ounce.
13
This contradicts Witness Haldi’s assertionthat worksharing increasesweight-related costs.
14
15
E.
IMPACT OF DATA FOR
15-16 OUNCE m
INTERVAL
16
Witness Tye critiques the statistical validity of Witness Daniel’s cost data for the highest
17
weight interval (15-16 ounces)in all classesof Standard(A) mail.z’ Witness Tye hypothesizes
18
that high unit costswithin the 15-16 ounce weight interval are the result of more tallies recorded
19
at this level than at the other heavy weight intervals.Y Therefore, WitnessTye concludesthat the
20
cost figures in the 15-16 ounceweight interval for all Standard(A) mail have greater support than
21
those of the other heavy weight increments in Witness Daniel’s study.
;
Tr. 30/14700.
Tr. 30/14701.
..
-26-
,
MOAA, ET AL.-RT-1
1
Whether or not the high unit costs in the 15-16 ounce weight interval are the result of more
2
tallies or another factor is irrelevant. The issueis whether or not the costscontainedin the 15-16
ounce weight interval of Witness Daniel’s study have a negative effect on her regression (i.e.,
reducesthe statisticalcorrelation). To test whether the 15-16ounceinterval is a statistical outlier
and significantly impacts the results of Witness Daniel’s analysis, I removed the data in her
analysis and recalculated the ECR regressionmodel for all shapescontained in LR-92. XZ’The
results obtainedby removing the 15-16ounceweight interval data com%mthat the data in the 1516 weight interval negatively impacts her cost study. A comparison of the results of the two
regressionsis shown in Table 5 below.
Table 5
Weight-Cost Relationship for ECR All ShapesData~emative
Rm
Cents Per Pound
m
--Coefficient
(1)
(2)
(3)
(4)
M’
CentsPer Piece
Coefficient
&pzr
(5)
Variatio$
(9)
(6)
(7)
(8)
1.3c
7.8C
6.5t
4.0
6.2
2.2
1.
USPSDaniel - All Data
18.4C
30.7c
43.oc
12.3.Z
-5.3c
2.
USPSDaniel 15-16oz.
Weight Interval Omitted
15.7
20.1
24.5
4.4
1.9
Sources: USPS-LR-l-92 Excel sheet:ECR All (detailed).
lf
Column (4) minus Column (4.
3
Column (7) minus Column (6).
22
Line 1 of Table 5 shows the upper-bound, lower-bound and regressioncoefftcient of ECR
23
mail cost on a per pound and cost per piece basis with the 15-16 ounce weight interval included
I+?’ From a statistical standpoint, any data point which lies more than four standard deviations away from a
regression lie created by regressing the data range without the suspect data can be considered an outlier.
-27-
MOAA, ET AL.-RT-1
1
as a data point.%’ Line 2 of Table 5 showsthe sametest results with the data for the 15-16 ounce
2
weight interval omitted from the regression. By omitting the 15-16 ounce weight interval, the
3
variation per pound is reducedfrom 12.3 centsper pound to 4.4 centsper pound (Table 5, Column
4
(5)). Similarly, when the data for the 15-16 ounceweight interval is eliminated, the variation per
5
piece decreasesfrom 6.5 cents per piece to 2.2 centsper piece (Table 5, Column (9)).
6
WitnessDaniel attemptsto temperthe impact of theseoutlying points on all Standard(A) mail
7
costsby combining the 15-16 ounce weight interval with other heavyweight intervals in her data
8
set (i.e., 13-14 ounce and 14-15 ounce intervals).fl’ Combining the heaviestweight-intervals as
9
WitnessDaniel hasdone may reducethe impact of the outlying data point but this adjustmentmay
10
also obscure significant information that the 15-16 ounce weight interval may contain. In my
11
opinion, when the data is properly analyzed(as I have done in Section V above) no reasonexists
12
to exclude (or otherwise aggregate)the data for the 15-16 ounce weight interval.
13
F. IMPACT OF WEIGHTED Dm
14
Witness Tye points out that Witness Daniel’s study provides equal statistical weight to each
15
weight increment while ignoring the varying volumes and weights in those increments For
16
example, Exhibit MOAA, ET AL.-RT-IB included in this testimony showsvolumesfor ECR mail
17
range from 13 million piecesin the 15 to 16 ounceweight interval up to 6.56 billion pieces in the
18
0 to 0.5 ounceweight interval. This reflectsa volume multiple of 502 (6.5 billion + 0.13 billion).
@’
A regression analysis provides statistically probable estimates for a value based on a specific confidence interval.
In this instance, I use a 95% confidence interval for estimates. The practical implications of the regression are
that in this instance, I can say with 95% confidence that the estimated cost of each variable (i.e., weight and
volume) lies between the upper-bound and the lower-bound and that the regression coefficient is the most likely
value.
4L’ Tr. 4/1343.
-28-
MOAA, ET AL.-RT-1
Sucha wide variancein a key variable (i.e., volume) will, in almost all cases,impact the results
of a regression analysis.
To test whether the variancesin weight and volume between the different weight intervals
affected ECR mail unit costs, I applied weighted regressionsto Witness Daniel’s data adjusting
for variances in weights and in volumes. QJThe results of Witness Daniel’s regressions and
regressionsweighted on pieces or pounds are shown in Table 6 below.
7
8
Table 6
dels%CR
Reeression
(1)
1.
2.
3.
USPSDaniel - All Data
USPSDaniel Interval 15-16oz. Omitted
WeightedRegressions
a. Weightedon Pieces
b. Weightedon Pounds
Cost Per
Pound (cent&
(2)
30.7c
20.1
11.0
17.9
sources:
Lines 1 and 2: USPS-LR-I-92Excel sheet: ECR All (detailed).
Lines 3: “PrescoaWorkpapersfor MOAA, ET ALRT-lxls” submitted
in support of this testimony.
21
The weighted regressionadjustedfor mail volume (Table 6, Line 3a) producesan estimated
22
cost of 11.0 cents per pound which is 64.2 percent lower than the Witness Daniel’s estimate of
23
30.7 centsper pound (Table 6, Line 1) and 45.2 percentper pound lessthan the results of utilizing
24
WitnessDaniel’s datawith the 15-16ounceweight interval omitted of 20.1 centsper pound (Table
25
6, Line 2). The weighted regression adjusted for total pounds produces an estimated cost per
42’
Weighted regression is a statistical procedure whereby data points are given statistical weights based up011a
causative factor. The effect is to apply greater value to those data points which have the greater infiuence.
-29-
MOAA,
ET AL.-RT-1
1
pound of 17.9 centsper pound (Table 6, Line 3b) which is 41.7 percentlessthan WitnessDaniel’s
2
original estimateand 10.9 percentlessthan WitnessDaniel’s estimatewith the 15-16ounceweight
3
interval omitted.
4
In my revised analysis, the implicit statistical weighting is basedon-
within a weight
5
interval versusthem
within a weight interval utilized by Witness Daniel. Exhibit MOAA,
6
ET AL.-RT-1B shows that, for ECR mail the 4.0 - 5.0 ounce weight interval contains the most
7
weight (848.9 million pounds)while the 15.0 - 16.0 ounceinterval containsthe leastweight (12.8
8
million pounds). Thus, the relationshipof the largestdata to smallestis a multiple of 66.3 (848.9
9
million + 12.8 million). The resultsof this reductionof the multiplier from 502 to 66.3 produces
10
a regressionthat is much closer to having equal statisticalweights amongstthe weight intervals.?’
11
12
G. IMPACT
OF THIN AND
NON-MAIL
13
Witness Tye criticizes Witness Daniel for utilizing data with a limited number of tallies for
14
a weight interval (i.e., “thin” tallies). Witness Haldi also criticizes Wimess Daniel for the
15
methodology she used to assign costs related to non-mail handling tallies to weight increments.
16
As I will discuss,both setsof criticisms are misplaced.
17
Witness Daniel’s methodology for estimating the unit costs divides total costs for a subclass
18
of mail by the total number of piecesfor that subclass. To arrive at total cost for each subclass,
19
WitnessDaniel summedthe aggregatecost for eachweight-interval within that subclass. Summing
20
data acrossall weight intervals implicitly assignsequal statistical weight to each weight-interval
9Y
A similar multiple also applies to Regular mail.
..
-3o-
MOAA,
ET AL.-RT-1
1
even though the number of tallies within the weight intervals may be dramatically different. In
2
other words, a weight interval which is based on a few tallies is assumedto be as equally
3
important as a weight interval basedon a large number of tallies.
4
The data utilized in my restatedregressionanalysishowever eliminates the reliance on those
5
weight intervals with tallies that haveless observationsthrough statisticalgrouping. a/ As shown
6
in Exhibit MOAA, ET AL.-RT-lD, the majority of the data points are grouped near the origin.
7
Included in this grouping are the data from the weight intervals with the lowest number of tallies
8
(i.e., weight intervals greaterthan 5 ounces). Statistically grouping the weight intervals with the
9
lowest number of tallies with those with largest number of tallies overcomesthe impact on the
10
11
overall analysis from those weight intervals which have fairly thin tallies.
In
contrastto Witness Daniel’s model, the weight interval in my analysis that is the furthest
12
from the origin is the 0 - 0.5 ounce weight interval which is shown as Data Point B in Exhibit
13
MOAA, ET AL.-RT-1D. As WitnessDaniel’s supportingdocumentationshows, the 0 - 1.0 ounce
14
weight interval hasthe largestnumber of tallies and the greatestaggregatecosts.’ By developing
15
the data as shown in my analysis, the data with the fewest tallies are grouped closely together.
16
This eliminates
17
and createsa more explanatory result.
the concern
raised by Witness Tye and Witness Haldi regarding
thinness
of tallies
441 There is a key distinction between gmuplng data as I do here and wmbiig
data as witness Daniel did in her
study. Witness Daniel combined and therefore lessened the number of data points (groups) in her study which
dropped the explanatory value of the data. My statistical grouping of the data did not eliminate or reduce the
number of weight intervals shldied.
2~’ Tr. 40344.
-311
2
3
4
5
6
MOM,
ET AL.-RT-1
Witness Haldi asserts that the USPS’ allocation of costs from non-mail handling tallies
systematically understatesweight-related costs. Witness Haldi states:
It seems completely inappropriate to use direct tallies from individual piecehandling operations to distribute to weight increment the costs associatedwith
some, if not all, of the not handling tallies. The effect of weight will be
systematically understated.ti’
7
He baseshis assertionon two underlying and unstatedassumptions:
8
9
1. That USPS equipment and personnel are always at full utilization and have no excess
capacity; and,
10~
2. All non-mail handling associatedcostsare driven by weight.
11
I discuss below why each of Witness Haldi’s assumptionsare erroneous and why the USPS’
12
allocation of non-handling tallies is correct.
13
Witness Haldi first assumesthat all USPS equipment and personnel have no excessor idle
14
capacity. In the hypotheticalexamplehe provides in his testimony, Witness Haldi assertsthat as
15
the weight of mail increases,the amountof equipmentrequired to handle that mail increasesand
16
in turn, the’cost of moving that equipment (both empty and loaded) also increases.g’Witness
17
Haldi’s assumptiononly holds true if all assets,including personnel, are at full capacity. If the
18
assetsare not at full capacity, then a larger amount of work may be performed without incurring
19
additional costs.
A4’ Tr. 32/15833.
w
Tr. 32/15832.
..
-32-
MOAA,
ET AL.-RT-1
1
A simple example will illustrate this point. Assume that one USPS employee within a
2
processingcenter was assignedto collect empty mail hampersand that the employee, when fully
3
utilized, could move one hundred (100) empty hampersper hourP8’If the processingcenter was
4
processingtwenty-five (25) empty hampersper hour, the one employee assignedto move empty
5
hamperswould be 25 percentutilized (25 empty hampers+ 100 hampercapacity). Next, assume
6
that the USPSreceivesan additional batch of 2 ounceflats that produce fifty (50) empty hampers
7
per hour. In total, the plant would now produce seventy-five (75) empty hampersper hour (25
8
for the first batch and 50 for the secondbatch). However, the plant would still need only the one
9
employee to move empty hampersbecausehe is only utilizing 75 percent of his capacity (75
10
empty hampers f 100 hamper capacity).
11
Witness Haldi’s theory assumesthat the plant would incur a threefold increasein non-mail
12
handling associatedcosts when the output for the empty hamper increased from 25 to 75.
13
However, as the exampleshows, becausethe plant was not fully utilized, it could absorbthe extra
14
output without incurring additional costs. Therefore, no additional non-mail handling costs are
15
incurred.
16
Witness Haldi next assumesthat all non-mail handling costs are driven by weight. This is
17
not true as shown by the USPS’ WitnessVan-Ty-Smith. In her testimony, WitnessVan-Ty-Smith
18
describessome exampleswhere a non-mail handling tally can occur:
W
For this example, I utilii an employee moving empty hampers for the reason that if he were tallied, he would
be recorded in a non-mail handling activity.
-33-
..
MOAA, ET AL.-RT-1
When not handling mail, the employeemay be observedto be between handlings
at the instant of observation, monitoring the operation of the equipment, on the
way to obtain empty equipment,on break, or performing incidental administrative
duties, to cite a few examples.*
5
Obviously many, if not all, of the instancescited by Witness Van-Ty-Smith are not affected
6
by the weight of a piece of mail (e.g., break time) thus not all non-mail handling costs are driven
7
by mail weight. Given that mail weight is not the driving cost factor as WitnessHaldi asserts,the
8
issuebecomesthe determination of the appropriatebasisto distribute non-mail handling tallies to
9
weight intervals for which no weight information is contained. In my opinion, the USPSaddresses
10
this issue correctly when it distributes the cost for non-mail handling tallies based on the
11
distribution of mail handling tallies.
12
H. SUMMARY
13
In summary, the criticisms raisedby WitnessesTye and Haldi related to the underlying dam
14
used in Witness Daniel’s study of the impact on costsdue to changesin weight (i.e., LR-92) have
15
beenaddressedin this proceeding. In my opinion, the underlying cost and weight data in LR-92
16
are reliable for use in evaluating if the USPS’ proposedrate structure, including the proposedper
17
pound portion of the rates for mail abovethe breakpoint.
W
Van-Ty-Smith, page 13.
.-
-341
VII.
COST DIFJXRJWIAL
MOAA, ET AL.-RT-1
FOR LETTERS AND FLATS
2
The USPS’ rate proposal for ECR mail includes a rate differential between piece-ratedw
3
letters and flats preparedat the high density or saturation level. To support the rate differential,
4
the USPSdevelopedthe mail processingand delivery costsassociatedwith each type of mail and
5
preparationlevel.“’ Table 7 below comparesthe proposedrate differential betweenletters and flats
6
with the cost differential calculatedby the USPS:
Table 7
Comparison of USPS Proposed Rate and
Cents Per
10
11
Item
(1)
&&J$
(3)
&g
(4)
ece
‘biff,,
(5)
12
13
14
Mall Processing and Delivery Costs 2’
a. High Density
b. Saturation
5.693
4.781
5.973
5.259
0.280
0.478
15
16
17
18
19
USPS Rate Proposal”
a. High Density
b. Saturation
15.2
14.3
15.4
14.8
0.2
0.5
Daniel, page 29.
Dane& page 29.
Moeller, page 28. Excludes destination entry diwxmts.
i:
22
As shown in Table 7 above, the USPScalculated a cost differential betweenletters and flats
23
of 0.280 centsper piece for high density mail and 0.478 centsper piece for saturationmail (Table
24
7, Line 1). In its rate proposal, the USPSproposeda rate differential betweenletters and flats of
t:
This reflects ECR mail that weighs less than 3.3 ounces, i.e., mail below the breakpoint.
The USPS’ unit costs for delivery were developed in USPS LR-I-95 (“LR-95”) and the unit costs for mail
processing were developed in USPS LR-I-96 (“LR-96”).
-35-
MOAA,
ET AL.-RT-1
1
0.2 centsper piece for high density mail and 0.5 centsper piece for saturationmail (Table 7, Line
2
2).
3
Witness Haldi assertsthat within ECR mail, the USPS“data systemssystematically overstate
4
the cost of letters while the cost of flats is correspondinglyunderstated.“” He baseshis testimony
5
on his belief that there exists a mismatch between:
6
7
(i)
the way the USPS’ Revenue,Piecesand Weight (“RPW”) systemrecordsrevenue,volume
and weight; and,
8
(ii)
the way that the IOCS developsmail processingand city carrier in-office costs.
9
According to Witness Haldi, this mismatch causesthe IOCS to misclassify “heavy” letters
10
(i.e., letter-shapedpiecesthat weigh in excessof 3.3 ounces)as nonletters for cost purposesand
11
causesthe IOCS to misclassify letter-shapedpieceswith detachedaddresslabels (“DAL”) as letters
12
instead of flats. Witness Haldi asserts that this “mismatch biases the letterkronletter cost
13
differentials usedfor ratemakingwithin all four Standard(A) subclasses.”ZI’ When he adjusts his
14
costsfor ECR mail for the claimed errors relatedto heavy weight letters, WitnessHaldi calculates
15
an additional cost differential betweenletters and flats of $0.291 centsper piece related to “heavy”
16
letters. Witness Haldi’s adjustment for DAL letters increaseshis calculation of the letter/flat
17
differential by 0.175 centsper piece. His total cost adjustmentequals0.466 cents per piece”’ He
18
then proposesto increasethe letter/flat differential by his claimed cost differential of 0.466 cents
3~ Tr. 32115765.
a’
Tr. 32115765.
.%’ Tr. 3205818.
..
-36-
MOAA. ET AL.-RT-1
1
per piece resulting in a letter/flat differential in rates of 0.7 centsper piece for high density mail
2
and 0.9 cents per piece for saturationmail w.
My analysis of Witness Haldi’s adjustmentto the USPS’ letter/flat differential is discussed
3
4
under the following topics:
A. Withdrawal of the DAL Adjustment
B. Witness Haldi’s Methodology for Heavy Letters
C. USPS’ Use of Correct Data
D. Summary
9
A.
-RAWAL
OF THE DAL ADJUSTMENT
10
At the time of the oral presentationof his testimony, Witness Haldi revised his calculations
11
to withdraw his endorsementof the letter/flat differential associatedwith DAL letters, i.e., 0.175
12
cents per piece of the total differential that he calculatedof 0.466 centsper piece?’ To reflect
13
this change,WitnessHaldi madetwo errors. First, Witness Haldi statedonly that Table 7 of his
14
testimony be corrected.ss He should have corrected the rates and percentagesin Table 2 and
15
Table 3 of his testimony which are also affectedby his withdrawn testimony.’ Second,in revising
16
his proposedrateslocated in Table 7 of his testimony, WitnessHaldi indicated that the piece rates
17
for flat high density and saturation mail should be reducedby 0.1 cent?’ However, in adjusting
18
the high-density and saturation ECR rates for flats without making an adjustment to any of the
&I’
XI’
g
Tr. 32/15772 and Tr. 32/15781. The saturation mail adjustment equals
0.9 cents per piece instead of 1.0 cents
per piece due to roundiig.
Tr. 32115854.
Tr. 32/15781 and Tr. 32/15782.
Tr. 32115855.
-37-
>
MOAA,
ET AL.-RT-1
1
other rates, WitnessHaldi’s rate structureis no longer revenueneutral. Stateddifferently, Witness
2
Haldi’s revised proposal will recover less in revenuesthan under his original proposal.
3
4
B.
WITNESS HALDI’S
METHODOLOGY
In order to correct for the assumedmismatch betweenthe USPS’ RPW and IOCS recording
of mail volumesand costs,WitnessHaldi proposesan adjustmentmethodologywhich redistributes
costsbetweenshape-basedcategories.59’However, WitnessHaldi’s methodologydoes not follow
8
the USPS’ procedureto identify shapebaseddifferences and contains a conceptualerror. These
9
shortcomingsin his methodology produce inaccurateresults and fail to correct the problem that
10
he is addressing.
11
1. Qmpa&Qn
12
The USPS identified the cost differences between letters and flats based on an analysis of
13
delivery costsas shown in LR-95 and mail processingcostsas shown in LR-96. Witness Haldi’s
14
analysis relies on Witness Daniel’s base data from the cost-weight study included in LR-92.m’
15
From the base data in LR-92, Witness Haldi concludesthat the overall letter/flat differential,
16
before any adjustmentfor all ECR mail, equals0.542 centsper piece. u Severalproblems exist
17
with the starting point in Witness Haldi’s methodology.
with USPS’ Pr&
18
First, the data in LR-92 includes more than mail processing and delivery costs (e.g.,
19
transportationcosts). Therefore, the letter/flat cost differential in WitnessHaldi’s analysisreflects
E,
a/
Tr. 3205815.
Tr. 32/15815.
Tr. 32/15818.
..
-38-
MOAA,
ET AL.-RT-1
1
cost componentsnot consideredin the USPS’ analysis or rate proposal. Second,the LR-92 data
2
was intended to show cost differencesby weight interval.
3
The data in LR-92 does not contain any of the mismatchthat Witness Haldi’s claims because
4
the detailed data for letter-shapedmail identifies both the costs and volumes for heavy letters”
5
Witness Haldi’s assumesthat 2.6 percent of the costs for Al letters is applicable to his “heavy”
6
letter adjustment. Then, he shifts this averagecoststo letters, but doesnot shift the corresponding
7
volumes for “heavy” letters that is also shown in LR-92. This procedureis, therefore, inaccurate.
8
Third, Witness Haldi’s basecost differential (before he begins his adjustments)for all mail
9
equals 0.542 centsper piece !z’ which is larger the cost differential calculated by the USPS for
10
either high density or saturation mail (seeLine 1 of Table 7 above). BecauseWitness Haldi has
11
utilized averagecostsfor letter and flat mail (at all weight intervals), his basestarting point reflects
12
the cost differences associatedwith mail other than high density and saturation mail, e.g., cost
13
differences due to the different mix of dropshipping The USPS’ analysis in LR-95 and LR96
14
specifically isolated the difference to only shape-relatedcost differences. For example, Witness
15
Daniel recognizedthat the cost differencecausedby dropshippinghad to be e1iminate.d“so that the
16
effect of finer depth of sort can be calculated in the absenceof dropshipping’@‘. Witness Haldi
17
has made no such adjustmentand, therefore, his cost difference betweenletters and flats includes
18
the averagecost difference due to dropshipping.
a’
;
Based ontheimproper
comparison
of the letter volume data utilized by Witness Daniel and the USPS’ Witness
Moeller, Witness Haldi implies that the percentage of letters above the breakpoint might be as high as 17.7
percent (Tr. 32/15814). Based on the LR-92 data, only 0.9 percent of the mail falls into the 3.5 ounce to 16.0
ounce range.
Tr. 32/15818, Lie 3 of Table A-2.
Daniel, page 28.
-39-
MOAA, ET AL.-RT-1
1
In summary, WitnessHaldi’s basictit costsfor his analysis are flawed. The starting point
2
for Witness Haldi’s analysis has not followed the USPS’ procedure. The basic cost differential
3
he develops of 0.542 cents per piece do not reflect the differences solely related to shapeand,
4
therefore, any adjustmentshe makesto the unit costs are invalid.
5
2. -
6
The purposeof examining the difference in costsbetween letters and flats is to determine if,
7
all other componentsare held constant, the shapeof a piece of mail causesa difference in its
8
cost.w WitnessHaldi’s methodologyutilizes the USPS’ volumes,but shifts his calculation of costs
9
for “heavy” letters from letters to flats. As shownbelow, his procedure is conceptually incorrect
10
and createsfurther misstatementof costs. If a mismatch in volume and cost data does occur, the
11
proper adjustment for purposesof determining the impact of shapeon costs is to reclassify the
12
piece count to match the correct shape-based
costs.A simple hypotheticalexampleof two mailings
13
shown in Table 8 below demonstratesthe impact from reclassifyingcostsrather than reclassifying
14
piece count. For this example, assumea mailing consistsof letter-shapedmail with 1,000 pieces
15
weighing less than 3.3 ouncesand 200 piecesof “heavy” letters that weigh more than 3.3 ounces
16
(seeLine la of Table 8 below). Next, assumea secondmailing of the samenumber of pieces,
17
consistingentirely of flat-shapedmail which has the samenumber of pieces and weight (Line lb
18
of Table 8 below). Further assumethat the costsfor the letter and flat mail are the same. Except
19
for the fact that one mailing is letter-shapedand the other is flat-shaped,the mailings are identical
$I’
As shown above, Witness Haldi’s analysis accounts for cost differences from nme than shape.
-4o-
MOAA,
ET AL.-RT-1
1
in every manner including the averagecost per piece (10 cents per piece for both letters and flats
2
as shown in Column (6) of Table 8).
-41-
MOM,
ET AL.-RT-1
Table 8
Hypothetical Example of Letter Flat Mismatch
..
.
. Volumes and Co&
-aI
:
3
4
5
Less
Than
3.3 O~ceS
p&.&
Q&t
(3)
(2)
Greater
Than
3.3 Ounces
Q&g
&g&g
(5)
(4)
Average
Cost
Per
&&
(‘5)
6
Assumed Mailing (Correct Data)
7
a.
Letters-ShapedMail
1004
$90
200
$30
$0. loo
8
b.
Flat-Shaped Mail
m
$9Q
m
m
k%QJK!
9
c.
Difference (Lla - Llb)
xxx
xxx
xxx
xxx
smoo
0
$30
$0.120
Assumed USPS Method of Recording
Mailings (Without Correction)
;:
12
13
14
E
,.
a.
Letter-Shaped Mail
loo0
$90
b.
Flat-Shaped Mail
m
sfl
em
m
$!QL?.&
c.
Difference (LZa - L2b)
xxx
xxx
xxx
$(0.034:
XXX
Haldi Adjustment to Correct
Mismatch Issue
17
a. Letter-Shaped Mail
loo0
$90
18
b.
Flat-Shaped Mail
m
m
s!x
&iQ
&Luu
19
20
c.
Difference (L3a - Ub)
xxx
xxx
xxx
$0.017
i:.
23
’
’
I
XXX
0
$0
$0.090
As recorded in RPW.
As recorded in IOCS.
[Column (3) + column (5)] f [Column (2) + Column (4)].
24
Line 2 of Table 8 illustrates the mismatch assumedby Witness Haldi without any correction
25
to classify the pieces for the two mailings.%’ Line 2 of Table 8 assumesthat the USPS’ RPW
6~’ Witness Daniel’s analysis of the cost difference between the mail processing and delivery costs of letters and flats
was constructed in a way that effectively reflects the unit costs on Lie 1 of Table 8.
-42-
MOAA,
ET AL.-RT-1
systemcategorizesletter-shapedpiecesover 3.3 ouncesas flats (Table 8, Column (4), Line 2b).
Line 2, Column (5) of Table 8 assumesthat the IOCS recognizesthe distinction between “heavy”
letters and flats by placing the costs into two separatecategories. Becausethe two reporting
systemstheoretically handle this heavy letter mail in a different manner, there is, according to
Haldi, a mismatchin piece count and cost. The allegedmismatch, under WitnessHaldi’s theory,
produces an overstatementin the cost per piece for letters and an understatementin the cost per
piece for flats. The total difference in this exampleequals 3.4 centsper piece (Column (6), Line
2c).
9
Witness Haldi’s proposed methodology for adjusting the USPS classification problem is to
10
move the shape-basedc,~& that are mismatched and leave the pi&e count in its misclassified
11
position. E’ Therefore, as shown in Line 3 of Table 8 above, Witness Haldi would reclassify the
12
costsfor all letter-shapedpiecesover 3.3 ouncesto the flat-shapedcost category (shifting the $30
13
in costsfrom Line 2a to Line 3b)pBI WitnessHaldi’s methodology results in an overstatementof
14
the averagecost of the flat-shapedpiecesand an understatementin the averagecost of the letter-
15
shapedpieces(Column (6), Line 3). Thus, WitnessHaldi’s approachdoesnot result in the correct
16
answer (i.e.; no difference in the averagecost per piece for the two samplemailings), but in fact,
17
now leads to an overstatementof flat-shapedaveragecost and an e
18
shapedmail. In other words, his adjustmenthas compensatedfor his perceived error (which in
a/
gsi
for the letter-
As discussed above, the data relied upon by Wimess Haldi did identify the letter volumes for the claimed “heavy”
letters so he could have moved the volumes for “heavy” letters.
Witness Haldi did not lmow the actual costs for these pieces but rather imputed the average costs of all letters
to the “heavy” letters (Tr. 32/15818).
-43-
MOM,
ET AL.-RT-1
1
actuality does not exist in the USPS’ methodology) in such a way as to causethe opposite effect
2
on the USPS’ data.
3
C.
‘USEOFCOBBECTDiQ+$
4
WitnessHaldi assertsthat this mismatchwas reflectedin the data in LR-92 utilized by Witness
5
Daniel. Even if Witness Daniel had relied on LR-92 to calculate the cost differential between
6
letters and flats, the data in LR-92 is not basedon the RPW systemas claimed by Witness Haldi.
7
In other words, the data sourcethat Witness Haldi claims reflects the mismatch was not used by
8
the USPS.
9
As Witness Daniel explained in her responseto interrogatory ADVONSPS-T28-1, she did
10
not rely on RPW volumes for the analysis in LR-92, but utilized PERMIT volume information.
11
Witness Daniel states:
12
13
14
The letter and nonletter volumesin USPS-LR-I-92are derived in USPSLR-I-102.
Thesevolumes are basedon the processingcategoryrecordedin PERMIT, which
should correspondto the DMM deftition of shape.@’
15
In responseto a questionposedby the Office of the ConsumerAdvocate (“OCA”) during oral
16
crossexaminationon April 12,2000, USPS’ WitnessMark E. Ramageconfii
17
hypothesizedby Witness Haldi related to LR-92 has been corrected in the USPS data:
18
19
20
21
that the mismatch
This questionis directed towards exploring the feasibility of adjusting IOCS data
so that it is consistentwith shapedefinition used for volume data for StandardA
letters. An alternative approach would be to produce volume estimates for
Standard A letters that are consistent with the IOCS shape definitions. My
@’
Tr. 4/1202,
-44
MOAA, ET AL.-RT-1
understanding is that witness Daniel employed this latter approach to ensure
consistency between the costs and volumes. See Tr. 40202. Since witness
Daniel relies on PERMIT volumes correspondingto the Domestic
Mail Manual
..
(“DMM”) shapedeftitions, she usesconsishapedefmons for her vo1ume
The IOCS shapedeftitions and the DMh4 shapedefinitions
and cost ea.
both define letter shapeaccording to the samephysical dimensionsof the piece.
SeeF-45, page 12-8, and CO50.2.0 of the DMM 55?’ (emphasisadded)
8
9
Witness Ramage’sstatementshows that any mismatch that had existed was corrected by the
USPS. Even if LR-92 was to be utilii
10
not), no adjustment is required.
11
D. SUMMARY
to calculatethe letter/flat cost differential (which it was
12
Witness Haldi claims that the USPSoverstatesthe cost of letters while understating the cost
13
of flats. His revisedcost differencebetweenlettersand flats equals0.297 centsper piece. Witness
14
Haldi has relied on the wrong basedata and, furthermore, his methodology contain a conceptual
15
error. These flaws demonstratethat his analysis is not valid. Therefore, the USPS’ results are
16
correct.
Ip’
Response of the United States Postal Service Witness Ramage to
Advocate During Cross-Examination filed 04/18/2tX0.
questionof
the Office of the Consumer
-45-
_A
1
VIII.
MOAA. ET AL.-RT-1
COST COVERAGE FOR STAM)ARD (A1 AND FIRST CI.,&S MAIL,
Witness Clifton proposesto reduce the cost coverageratios of First Class and First Class
Presort mail by 1.3 percentagepoints and 7.0 percentagepoints respectively’l’ To balance his
proposedreductionsin the Fist Classand Fist ClassPresortcost coverageratios, Witness Clifton
5
proposes to increase the cost coverageratio for Standard (A) Regular mail by 9.3 percentage
6
points (from the USPS’ proposed coverage 132.9 percent to 142.2 percent) and the ratio for
7
Standard (A) ECR mail by 5.6 percentagepoints (from the USPS’ proposed coverage ratio of
8
208.8 percent to 214.4 percent). Witness Clifton baseshis changes in cost coverage on his
9
assertionthat the cost coveragefor First Class mail has becomehighly discriminatory relative to
10
Standard(A) mail.Y’
11
This section of my testimony discussesWitness Clifton’s justification of his changesto the
12
cost coverages for Fist Class and Standard (A) mail and explains why his proposal is not
13
supportedby the data in this proceedingor by PRC precedent.
14
My review is discussedunder the following topics:
15
A. Changesin First Class Presort Cost Coverage
16
B. Changesin ECR Cost Coverage
17
C. Contribution to Institutional Costs
18
D. Summary
zl’
zz’
Tr. 26112457.
Tr. 26ll2463.
.
-46-
_.
1
2
A. CHANGES
3
4
MOAA,
ET AL.-RT-1
IN FIRST CLASS
WitnessClifton assertsthat First ClassPresortmail is being unfairly burdenedin its allocation
of institutional co~ts.~ In his testimony, Witness Clifton states:
5
6
7
8
9
While First Class workshared mail is supposedto be part of a single Fist Class
letters subclass,it doesappearunmistakably that in the growing disparatetrends
betweencost coveragesfor single piece versusworkshared mail in the allocation
of institutional costs, workshared mail is being singled out in an arbitrary and
almost punitive way.N
10
In support of his assertion,Witness Clifton provides a comparisonof implicit cost coverage
11
ratios for First Class and Standard(A) mail for the years 1994-1999.=’While it is clear that the
12
cost coveragefor First Class Presort mail has increasedin the last five years, Witness Clifton’s
13
comparison does not demonstratethe causeof the increasein the coverageratios.
14
By definition, cost coveragefor a given subclassof mail is the ratio of revenuesto volume
15
variable costs for that subclassof mail. Increasesin cost coverages,therefore, can occur either
16
through an increasein revenues,a decreasein costs, or a combination of both. Table 9 below
17
summarizes, for 1994 and the Test Year After Pates (“TYAR”) analysis presentedby the USPS
18
in this proceeding, the average revenue, costs and cost coverage ratio for First Class Presort
19
mailz’
;;
LV
761
Tr. 26/12460.
Tr. 26112460.
Tr. 26112459.
I am aware of the USPS’ supplemental testimony regarding the update of costs to reflect 1999 base year data.
For comparabiiity with Wi&ess Clifton, I have continued to utilize the same data as presented by Wimess
Clifton.
-47-
_.
MOM,
ET AL.-RT-1
Table 9
Comparison of Changes in
:
3
__ .
3
Percent
4
5
m”
(1)
(2)
(3)
LzclMn&
(4)
6
1,
Revenue-Cents Per Piece
26.1
28.15
7.9%
7
2.
Volume Variable Costs--Cents Per Piece
11.7
L@&?
=fLl%2
8
3.
Contribution -Cents Per Piece
14.4
17.47
21.3%
9
4.
Cost Coverage Ratio (Ll + L2)
223%
264%
18.4%
11
21
Fiscal Year 1994 Cost and Revenue Analysis
10
USPS-T-32,
ExhibitUSPS-32B,
pagesl-2.
21 [Column(3)I Column(2)] - 1.
i:.
13
14
Between 1994 and the TYAR, the cost coverageratio for First Class Presort mail rose 18.4
15
percentfrom 223 percentB’ to 264 percent(Table 9, Line 4). While revenueper piece increased
16
by 7.9 percent during the study period (Line 1, Column (4)), the volume variable costs decreased
17
at a rate of 8.7 percent (Line 2, Column (4)). In other words, approximately 50 percent of the
18
increasein the cost coveragefor Fist Class Presort mail is due to decreasedcosts.
19
The PRC has statedin previousdecisionsthat increasesin cost coveragedue to decreasesin
20
costs are not a sign of an unjust burden placed on a particular subclassof mail. In Docket No.
21
MC95-1. the PRC stated:
=’
The data in Table 9 for 1994 is based on the USPS’ CRA. Witness Clifton’s Table 12 shows a coverage rati
of 219 percent.
-48-
MOAA, ET AL.-RT-1
‘I.. .in every situation in which somemail allows the PostalServiceto avoid costs,
the implicit cost coveragefor that mail will be higher than the implicit coverage
for otherwise similar mail. The Commission believesthat this is just.lm’
1
2
3
The logic of the PRC’s decision in Docket No. MC95-1 applies in this proceeding. The
increasein Fist Class Presort mail cost coveragehas come about in large part, due to the lower
costs. Therefore, becausethe cost coveragehas increasedfor First Class Presort mail, this does
not mean, in light of the PRC decision in Docket No. MC95-1, that this particular subclassof mail
8
is being singled out for discriminatory rate increases.
9
B. CHANGES
10
11
IN ECR COST COVERAGE
Witness Clifton portrays the USPS’ proposedchangesin rates as unfairly placing a greater
burden on First Class mailers who have investedheavily in technology to reduce cost:
12
13
14
15
This (disparatetrends betweencost coveragesfor single piece versus workshared
mail) is unfair, inequitable, and discriminatory treatment towards the mailers
whose substantialinvestmentsand ongoing dedicationnow move 45 billion pieces
of Fist Class Mail through automatedprocessingtechnology annually?’
16
An examination of data for another subclassof mail (i.e., ECR mail) that relies heavily on
17
technology reveals that the contributions from First Class Presort mail arc not excessive.
18
19
Table 10 below, which follows the sameformat as Table 9, summarizesthe average revenue
per piece, costs per piece and contribution per piece for ECR mail for 1994 and TYAR.
z
Docket No. MC95-1 decision,
Tr. 26/12460.
pages111-28.
-49-
MOAA.
ET AL.-RT-1
Table 10
Q
3
4
Item
(1)
5
Revenue--Cents Per Piece
TyBB zi
Percent
&&f
(2)
(3)
(4)
13.2
15.72
19.1%
m
?/
6
,_
Volume Variable Costs--Cents Per Piece
Q
L-i3
lk?.a.
7
8.
Contribution -Cents Per Piece
6.9
8.2
18.8%
8
.
Cost Coverage Ratio (Ll i L2)
210%
209%
(-)l%
’
’
/
Fiscal Year 1994 Cost and Revenue Analysis
USPS-T-32, Exhibit USPS-32B; USPS-LR-I-166, WP 1, page 24.
[Column (3) I Column (2)1 - 1.
9
::
12
13
A comparisonof Table 9 and Table 10 illustratestwo key points. Becausethe revenuesand
14
costs changed at approximately the same level, the coverage ratio for ECR mail decreased1
15
percent between 1994 and TYAR. However, ECR mail has a much larger increasein revenues
16
than that borne by First ClassPresortmail. ECR revenueper piece equaled 13.2 cents per piece
17
in 1994 (Table 10, Line 1). Under the USPS proposed rate structure, ECR revenueper piece
18
increasesto 15.7 centsper piece for TYAR (Table 10, Column (3), Line l), an increaseof 19.1
19
percent (Table 10, Column (4). Line 1). In contrast to the 19.1 percent increasefor ECR mail,
20
First Class Presortmail revenueper piece increased7.9 percentover the sametime period (Table
21
9, Column (4), Line l), a difference of 11.2 percent.
22
Second,the unit contribution for ECR mail increasedby 18.8 percent,over the 1994to TYAR
23
time period (Table 10, Column (4)). As Table 9 shows, the unit contribution for First Class
24
Presort mail increased21.3 percent between 1994 and TYAR (Table 9, Column (4), Line 3).
..
-5o-
..
MOAA,
ET AL.-RT-1
1
Therefore, the changein the contribution is approximatelythe samefor First Classand ECR mail.
2
The fact that ECR mail will maintain a percentagechangein unit contribution essentially equal to
that of Fist ClassPresort mail while also seeinga substantialincreasein unit revenuecompared
to First Class Presort mail is a clear indication that the First Class Presort mail is not receiving
discriminatory treatment when comparedto ECR mail. 8p,
6
C. CONTRIBUTION
TO INSTUUIlONAL
COSTS
7
WitnessClifton assertsthat the cost coverageratios for First Classmail and Standard(A) mail
8
should be adjusted becauseFirst Class mailers contribute an excessiveportion to the USPS
9
institutional costs. He states:
10
11
12
13
There is an additional reason why the Commission should adjust the cost
coveragesalong the lines I suggest. The contribution the Fist Class letter mail
subclassmakesto USPSinstitntional costshas simply gotten out of hand over the
199os.U’
14
Although First Classmail has seenan increasein contribution, Standard(A) Regular mail has
15
seenan evengreaterincreaseto contribution. Table 11 below, which is basedTable 13 of Witness
16
Clifton comparesthe absoluteand relative changein contributionsto institutional costsfrom First
17
Class letter mail and Standard (A) Regular mail for the time period 1994 to 199F’. The First
18
Classletter mail subclasshas showna greaterabsolutechangein contribution over this time period
19
than has Standard (A) Regular mail. This absolutechange is to be expectedgiven the greater
8a’
Witness Tye also assertsthat the proposed change in rates unfairly places a greater burden on First Class mailers
(Tr. 30/14731). Lie Witness Clifton, Witness Tye does not look at the relative change in unit contribution and
therefore does not recognize that ECR and Fist Class mail have received an essentially equal relative increase
in unit contribution.
s?’ Tr. 26/12460.
SZ’ This is the same period Witness Clifton utiliies in Table 13 of his testimony.
.
-51-
..
MOAA,
ET AL.-RT-1
1
volume in Fist Class mail as well as Standard (A) mailers greater use of destination entry
2
discounts which lower overall revenueand costs.
Table 11
p
5
6
7
Casts As Shown bv Clifton
Mail
ClaSS
(1)
8
1.
First Class Letter
9
2.
Standard (A) Regular
10
3.
Total Mail Service
11
12
13
IL/
Cl&m.
lw
(2)
ontrtbu&ztt 090)”
.l.B
(3)
% change
1994-1999”
(4)
45.8%
$11,410
$16,640
1,211
2,084
72.1
17,284
24,265
40.4
Table 13 (Tr. 26/12461).
14
As shown in Table 11 above, First Class letter mail has seen a 45.8 percent (Table 11,
15
Column (4), Line 1) increase in its contribution to institutional costs for the 1994-1999 time
16
period. This is close to the USPSoverall 40.4 percentchange(Column (4), Line 3) for the same
17
time span. In contrast, Standard(A) Regular has seena 72.1 percent increasein its contribution
18
(Table 11, Column (4), Line 2), well abovethe overall USPSaverage. To assertthat Fist Class
19
letter mailers haveseendiscriminatory increasesin relative contribution over the 1994-1999time
20
period disregardsthe large increaseincurred by Standard(A) Regular mailers.
21
D. HJMMARY
22
Witness Clifton infers throughout his testimony that First Class mail carries too much of an
23
institutional cost burden and, therefore, cost coveragesshould be adjusted. He recommendsthat
-52-
MOAA,
ET AL.-RT-1
1
First Class single piece and Fist Class Presort mail cost coveragesshould be lowered and
2
Standard (A) Regular and ECR mail coveragesbe raised. An examination of the argumentshe
3
uses to support his proposed changesreveals that he has disregarded past PRC decisions and
4
utilized data which ignores increasesin revenueand contribution for Standard(A) mail. Based
5
on previous PRC decisionsand data for Standard(A) mail for the sametime period as utilized by
6
Witness Clifton, no basis exists to increasethe coverageratio for Standard (A) mail above the
7
USPS’ proposed level.
Exhibit MOAA, et al.-RT-1A
Page 1 of 1
USPS’ Costs and Volumes by Weiaht Increments -- Reeular
~hputs used by Ms. Daniel
Weight
Number of
Pieces per
Increment II
(2)
Weight
Average
Weigbt
Per Piece
Inputs for Rev&d Regression
Average
Pieces
Per Pound 4/
(5)
Average
cost
Per Piece 21
(6)
(7)
Average Cost
Per Pound 5/
(8)
(Pounds111
Total Cost
Jl,OOO’s~I/
@uwes) 2/
(3)
(4)
1.
0.0 to 0.5
8,747,091,966
184,280,580
$1,081,748
0.34
$0.124
47.47
$5.870
2.
0.5 to 1.0
11,404,201,293
519,125,736
$1,455,419
0.73
$0.128
21.97
$2.804
3.
1.0 to 1.5
4,792,879,103
367,132,978
$731,699
1.23
$0.153
13.05
$1.993
4.
1.5 to 2.0
2,988,638,371
322,254,136
$554,328
1.73
$0.185
9.27
51.720
5.
2.oto 2.5
2,103,443,012
295,055,711
$403,113
2.24
$0.192
7.13
$1.366
6.
2.5 to 3.0
2,549,930,575
441,438,182
$438,169
2.77
$0.172
5.78
$0.993
7.
3.0to 3.5
2,498,208,591
502,568,111
$428,771
3.22
$0.172
4.97
$0.853
8.
3.5 to 4.0
1,523,657,694
356,425,916
$492,101
3.74
50.323
4.27
$1.381
9.
4.oto 5.0
2,192,214,612
608,987,097
$346,338
4.44
$0.158
3.60
$0.569
10.
5.Oto 6.0
1,253,983,750
426,670,168
$244,717
5.44
50.195
2.94
so.574
11. 6.0 to 7.0
722,093,403
291,671,566
$170,430
6.46
50.236
2.48
SO.584
12.
7.0to 8.0
486,188,828
226,985,241
$184,911
7.47
$0.380
2.14
$0.815
13.
8.Oto 9.0
333,826,177
176,730,047
$99,212
8.47
$0.297
1.89
50.561
14.
9.0 to 10.0
244,795,395
145,275,303
$109,578
9.50
50.448
1.69
$0.754
15. 10.0 to 11.0
2463682,929
162,410,751
$100,045
10.53
$0.406
1.52
50.616
16. 11.0 to 12.0
202,579,432
145,515,879
$100,442
11.49
$0.496
1.39
50.690
17. 12.oto 13.0
216,130,522
169,177,817
$103,269
12.52
50.478
1.28
50.610
18. 13.oto 14.0
133,968,247
112,813,257
$81,984
13.47
50.612
1.19
50.727
19. 14 .oto 15.0
85,577,382
77,255,918
$60,035
14.44
$0.702
1.11
$0.777
20. 15.Ota 16.0
57,681,913
55,765,086
$75,061
15.47
$1.301
1.03
$1.346
I/
USPS-LR-I-92 sheet Regular all (detailed)
z/
Column (3) + Column (2) x 16 ounces
g
(column (4) x 1,000) + Column (2)
41 Column(2)+
y
Colllmn(3)
(Column (4) x 1,000) + column (3)
Exhibit MOAA, et al.-RT-1B
Page 1 of 1
USPS’ Costs and Volumes bv Weight Increments -- ECR
Inputs used by Ms. Daniel
Average
weight
Average
Per Piece
cost
J0unce.d ?.I Per Piece 31
(6)
(5)
Inputs for Revised Regression
Average
Pieces
Per Pound 41
(7)
Average Cost
Per Pound 5/
F-9
$0.058
48.53
$2.838
0.73
50.072
21.79
$1.560
$212,642
1.28
50.076
12.53
$0.954
323,500,191
5196,100
1.78
50.068
8.97
50.606
3,548,811,635
509,565,386
$218,277
2.30
$0.062
6.96
$0.428
2.5 to 3.0
2,960,135,421
519,324,061
$204,524
2.81
50.069
5.70
50.394
7.
3.0 to 3.5
1,875,267,345
385,734,533
$157,908
3.29
50.084
4.86
$0.409
8.
3.5 to 4.0
1,549,324,284
372,534,646
5158,875
3.85
50.103
4.16
$0.426
9.
4.0 to 5.0
2,977,269,831
848,935,134
$213,855
4.56
$0.072
3.51
$0.252
10.
5.0 to 6.0
1,342,660,886
464,229,728
5114,417
5.53
50.085
2.89
$0.246
11. 6.0 to 7.0
699,669,330
288,375,650
565,932
6.59
50.094
2.43
50.229
12.
7.0 to 8.0
371,958,415
176,937,461
542,400
7.61
50.114
2.10
50.240
13.
8.0 to 9.0
201,513,104
109,179,206
$26,672
8.67
50.132
1.85
50.244
14. 9.0 to 10.0
78,920,017
47,711,180
515,622
9.67
50.198
1.65
$0.327
15. 10.0 to 11.0
74,474x482
49,701,200
510,533
10.68
50.141
1.50
$0.212
16. 11.0 to 12.0
33,831,994
24,918,961
$8,264
11.78
50.244
1.36
$0.332
17. 12.0 to 13.0
32,205,634
25,756,359
$5,828
12.80
50.181
1.25
50.226
18. 13.0 to 14.0
25,434,174
21,883,581
55,093
13.77
50.200
1.16
$0.233
19. 14 .oto 15.0
17,179,749
16,012,076
54,460
14.91
$0.260
1.07
$0.279
20. 15.0 to 16.0
13,060,565
12,809,676
57,852
15.69
$0.601
1.02
50.613
Weight
Increments
(ounfes)
(1)
Number of
Pieces per
Increment 11
(2)
1.
0.0 to 0.5
6,567,978,563
135,341,727
5384,125
0.33
2.
0.5 to 1.0
5,568,422,818
255,493,988
$398,526
3.
1.0 to 1.5
2,790,971,660
222,825,324
4.
1.5 to 2.0
2,901,427,528
5.
2.0 to 2.5
6.
Weight
(pounds) I/
(3)
I/
USPS-LR-I-92 sheet ECR all (detailed)
g/
GAunm (3) + Column (2) x 16 ounces
y
(colllmn (4) x 1,000) + Column (2)
A/ Column (2) + Column (3)
5/
(Column (4) x 1,000) + Column (3)
Total Cost
p.ooo’s~ I/
(4)
Exhibit MOAA, et al.-RT-1C
Page 1 of 2
Dollars
Per
Pound
Restated Regression Utilizing
Cost Per Poulid and Pieces Per Pound
(Regular)
$6.00 -
$5.00 -
$4.00 -
$3.00 -
$2.00 -
$1.00 -
$0.00
I
1
8
1
I
I
0
10
20
30
40
50
Number of Pieces Per Pound
Source: MOAA. et al-RT- Ids
-
:
Exhibit MOAA, et al.-BT-1C
Page 2 of 2
D;r
Restated Regression Utilizing
Cost Per Pound and Pieces Per Pound
@CR)
$3.00
) DataPointB
Pound
$2.50
$2.00
/
10
0
20
30
Number of Pieces Per Pound
Source: MOM, et al-RT-l.xls
.
-.
. .
40
50
1
Exhibit MOAA, et al.-RT-1D
Page 1 of 1
USPS Witness Daniel’s Estimated
Standard (A) ECR Unit Cost Per Piece
Cents
Per
Piece
70
1
60 -
] DataPointA
1H
50 40 -
0123456789
10
Weight Per Piece (ounces)
Source: USPS-LR-92
-
.I
-
-
_.
_
,,.
-
11
12
13
14
15
16
Exhibit MOAA, et al.-RT-1E
Page 1 of 1
Unit Cost Line Based on Restated Regression - Remlar
Cents
70
Per
1
Piece
60 50 40 -
Standard (A) Regular Unit Costs = 11.1 per piece +
(52.5 6 per pound + 16 ounces x avg. weight per piece)
0
I
I
I
0
1
2
3
I
I
I
I
4
5
6
7
8
I
I
I
I
I
I
I
I
9
10
11
12
13
14
15
16
Weight Per Piece (ounces)
Source: MOM,
.
et al-RT-1.~1~
_I
.,,
_--_.._.I
Exhibit MOAA, et al.-RT-1F
Page 1 of 1
Unit Cost Line Based on Restated Regression -- ECR
Cents
Per
25 1
Standard (A) ECR Unit Costs = 5.6 6 per piece +
(17.6 $ per pound + 16 ounces x avg. weight per piece)
0
1
2
3
4
5
6
7
8
9
Weight Per Piece (ounces)
Source: MOAA, ef al-RT-l.xls
10
11
12
13
14
15
16